Online car-hailing order dispatching method integrating order priority ranking

Abstract

The invention discloses an online car-hailing order dispatching method fusing order priority ranking. The method comprises the following steps: step 1, extracting to-be-matched orders in a current system order pool; step 2, generating an order priority queue based on a preset rule; step 3, constructing a vehicle-order weighted bipartite graph matching model; step 4, solving the priority of the KM algorithm step by step; and step 5, executing rolling updating of the order sending system. According to the method, matching fairness is improved, controllability and elastic scheduling capability are enhanced, algorithm expandability is improved, expandability and adaptability of a model are enhanced, and the problems that in the prior art, order priority is neglected, a scheduling mechanism is rigid, and fairness and efficiency are difficult to consider at the same time are effectively solved. And the service quality and the user satisfaction of the online car-hailing platform are improved.

Description

Technical Field

[0001] The present invention relates to a method for dispatching online ride-hailing orders, and in particular to a method for dispatching online ride-hailing orders that integrates order priority sorting. Background Art

[0002] With the development of intelligent urban transportation, mobile internet-based ride-hailing platforms have become an important means of transportation for urban residents. These platforms typically utilize an order dispatching system to match users with available vehicles in the vicinity after they initiate a travel request. To improve vehicle utilization and user satisfaction, dispatching systems often employ order-vehicle matching algorithms to perform real-time matching between the order pool and the pool of available vehicles.

[0003] Currently, mainstream platforms mostly use a greedy dispatching strategy centered around distance. For example, they assign orders to the nearest available vehicle, or sort orders based on a rule-based weighting function (such as distance + driver rating) and then match them sequentially. This strategy offers certain computational efficiency and interpretability, but also has significant limitations:

[0004] Insufficient fairness: The system struggles to take into account the waiting times of all users, and users with long waiting times may be continuously ignored.

[0005] Inefficient resource utilization: A matching mechanism based solely on distance may ignore the overall matching quality of the system;

[0006] Unable to dynamically adjust matching priorities: Lack of differentiated consideration of dimensions such as user level and order value.

[0007] Current online ride-hailing platforms generally adopt greedy dispatch strategies based on geographic distance or time cost, or introduce weighted bipartite graph matching algorithms (such as the KM algorithm) to optimize overall dispatch costs. However, these methods have the following problems in practical applications:

[0008] 1. Ignoring order priority differences: Existing algorithms typically treat all orders as equally important, without considering factors such as order waiting time, user level, and historical evaluation. This can result in high-priority orders being "preempted" by low-priority orders, impacting user experience.

[0009] 2. Lack of dynamic scheduling mechanism: During peak hours or when resources are tight, the existing system is difficult to dynamically adjust based on real-time changes in order and vehicle status, resulting in reduced scheduling efficiency.

[0010] It is difficult to strike a balance between fairness and efficiency: while pursuing overall scheduling efficiency, existing methods often ignore the fairness needs of individual users and cannot ensure that all users receive reasonable service responses.

[0011] The prior art closest to the present invention is an order-vehicle weighted matching scheduling method based on the KM algorithm. The core idea of ​​this method is as follows:

[0012] Construct a weighted bipartite graph of all the orders to be matched and the idle vehicles in the current system;

[0013] The matching edge weight is usually composed of comprehensive indicators such as pick-up distance, estimated time, and service cost;

[0014] The KM algorithm is used to solve the optimal matching of the bipartite graph and obtain the vehicle-order allocation solution;

[0015] The matching results are directly used for order execution and are updated periodically.

[0016] This type of method is theoretically globally optimal and can improve the overall efficiency of the scheduling system. However, it has significant shortcomings in order priority processing capabilities:

[0017] This matching method treats all orders as “equivalent tasks” and ignores heterogeneous factors such as user waiting time, order value, and user reputation;

[0018] During peak hours or when resources are scarce, high-priority orders may still be "preempted" by low-priority orders, affecting user experience;

[0019] Lack of controllability and flexibility, and inability to flexibly adjust scheduling strategies to adapt to different business demand scenarios.

[0020] Therefore, existing technologies are difficult to meet the scheduling optimization goal of "balancing fairness and efficiency", especially in actual operating scenarios where order demand is highly dynamic and user structure is complex, there is still much room for improvement in the service fairness and satisfaction of its matching results. Summary of the Invention

[0021] Purpose of the Invention: The purpose of this invention is to provide a method for dispatching ride-hailing orders that integrates order priority sorting. This method addresses the issue of how to balance fairness in processing user travel orders with system matching efficiency during the dispatch process, ensuring that high-priority orders (e.g., orders with long waiting times or high user levels) are responded to first, thereby improving overall dispatch quality and user satisfaction.

[0022] This online car-hailing dispatching method that integrates order priority sorting aims to: 1. Introduce an order priority mechanism: Dynamically evaluate the priority of orders based on factors such as the waiting time of the order, user level, and historical evaluation to ensure that high-priority orders are served first. 2. Combine the weighted bipartite graph matching model: When constructing the vehicle-order matching model, comprehensively consider factors such as pick-up distance, time cost, and energy consumption to optimize the overall dispatching cost. 3. Implement a rolling update scheduling mechanism: By periodically updating the order and vehicle status, dynamically adjust the dispatching strategy to improve the system's responsiveness and dispatching efficiency. 4. Take into account both fairness and efficiency: While ensuring that high-priority orders are responded to first, optimize the overall dispatching cost to achieve a balance between fairness and efficiency.

[0023] The above-mentioned technology effectively solves the problems existing in the existing technology, such as neglect of order priority, rigid scheduling mechanism, and difficulty in balancing fairness and efficiency, thereby improving the service quality and user satisfaction of the online ride-hailing platform.

[0024] Technical solution: The method for dispatching online ride-hailing orders integrated with order priority sorting includes the following steps:

[0025] Step 1: Extract the orders to be matched from the current system order pool;

[0026] Step 2: Generate order priority queue based on preset rules;

[0027] Step 3: Build a vehicle-order weighted bipartite graph matching model;

[0028] Step 4: KM algorithm priority is solved step by step;

[0029] Step 5: Rolling update of the dispatch system is executed.

[0030] Furthermore, the method of step one is as follows: a real-time order management module is set up inside the dispatch system, and the module is connected to the order database of the online car-hailing platform through standard data interface technology, from which all user orders currently in the waiting state are extracted periodically or in real time, and a set of orders to be matched is formed.

[0031] Furthermore, the order information is stored in a high-speed cache database within the dispatch system in a structured data format.

[0032] Furthermore, the method of step 2: the order dispatching system obtains the set of orders to be matched To evaluate and sort the priority of each order in the dispatch system, an order priority evaluation module is set up inside the dispatch system. This module evaluates each order based on pre-defined rules and assigns corresponding priority labels.

[0033] Furthermore, the priority evaluation rules include: order waiting time, order estimated cost, user historical evaluation level, and maximum waiting time acceptable to the user.

[0034] Furthermore, the order priority classification method is performed based on the order waiting time: In order to clarify the specific order priority classification process, a time granularity control parameter Δt is introduced f , which is used to segment the waiting time of orders and divide the order queue according to the following method:

[0035]

[0036] Where, Indicates order o i Current waiting time, The maximum waiting time for an order allowed by the user, according to the above method, the order collection Divide into K subsets, each subset is assigned a priority label It can be seen that the larger the priority tag value of the order subset, the longer the waiting time of the orders in the subset, the higher the priority, and the priority will be given priority in the subsequent order dispatch process. Finally, the system will prioritize the processing and matching based on the subset priority tag value ρ. k Sort all order subsets to form a complete order priority queue The priority of the head subset of the queue is the highest, and the priority of the tail subset is the lowest. The generation process of the above order priority queue is automatically completed by the order priority evaluation module in the dispatch system, and the results are stored in the system memory database or cache module for rapid call by subsequent vehicle order optimization matching decisions.

[0037] Furthermore, the method of step 3 is as follows: at each scheduling time t, the set of vehicles currently available for receiving orders is recorded as The set of user orders to be served is denoted as O t ={o1, o2, ..., o n}, to model the matching relationship between vehicles and orders, a weighted bipartite graph G = (X∪Y, E) is constructed, where the node set X = {x1, x2, ..., x m} corresponds to the available vehicles, the node set Y = {y1, y2, ..., y n}Corresponding to the order to be matched, the edge set E in the figure represents the reachable or serviceable relationship between the vehicle and the order. As shown in the figure, each edge in the bipartite graph is accompanied by a matching cost weight like The weight represents the comprehensive cost required to assign vehicle v to order o.

[0038] In this invention, the matching cost of each edge is Consider the following three factors:

[0039] Pick-up distance The distance traveled by vehicle v to the location of the passenger of order o (unit: kilometers);

[0040] Pick-up time The time required for the vehicle to reach the order passenger's location (unit: minutes);

[0041] Energy consumption The amount of electricity required by the vehicle to complete the pick-up process (unit: kWh),

[0042] The edge weight is defined as:

[0043]

[0044] Among them, α1, β1, and γ1 are the weight coefficients of the three indicators, which are set according to the platform strategy or business preference to balance the impact of different factors on the total cost.

[0045] Therefore, the specific objective function is to minimize the overall order-taking cost of the fleet:

[0046]

[0047] in is a decision variable, indicating whether the order o is served by vehicle v at time t, and takes the value of 0 or 1.

[0048] To ensure the feasibility of the matching scheme, this model needs to meet the following constraints:

[0049] (1) Each vehicle can only serve at most one order at a time;

[0050]

[0051] (2) Each order can only be served by one vehicle at a time;

[0052]

[0053] (3) If the order waiting time exceeds the maximum allowed value, the match will be automatically cancelled, where M large is a sufficiently large number used to construct the conditional judgment of the big M method.

[0054]

[0055] (4) The vehicle battery power must meet the requirement that the remaining mileage after completing the order service is not less than the preset minimum mileage requirement to ensure that the vehicle can reach the nearest charging station, where M large is a sufficiently large number used to construct the conditional judgment of the large M method; dmin The minimum driving distance that the vehicle must maintain after completing the order. is the pick-up mileage, d o is the mileage between the start and end point of the order, is the current remaining driving range of vehicle v,

[0056]

[0057] (5) Binary constraints on decision variables:

[0058]

[0059] Furthermore, after completing the vehicle-order weighted bipartite graph, the generated order priority queue is combined The Kuhn-Munkres algorithm (KM algorithm) is used to solve the order subsets of different priorities in turn to achieve the optimal matching that takes into account both fairness and efficiency of order dispatching.

[0060] Furthermore, the dispatch system generates order priority queues based on the generated order priority queues. For each priority subset The KM algorithm is called in order from high priority, long waiting time, to low priority for matching. The specific process is as follows:

[0061] (1) Initialize resource collection:

[0062] Currently unassigned vehicle collection:

[0063] The current priority level index is processed from k=K to k=0 in sequence

[0064] (2) Priority subgraph extraction:

[0065] For the current priority order subset Select vehicles that can accept orders from the remaining vehicle set and construct a local subgraph The edge weight is defined in step 3

[0066] (3) KM algorithm call:

[0067] In each priority subgraph G k On the top, call the KM algorithm to solve the minimum cost matching (or maximum negative weight matching)

[0068] (4) Result update:

[0069] The current matching result M k Merge into the total matching result set M←M∪M k

[0070] Remove the vehicle that has participated in the match from the remaining vehicle set: V t,remain ←V t,remain \Dom(M k )

[0071] (5) Iterate layer by layer until the lowest priority is processed

[0072] Through the above mechanism, high-priority orders are given priority in resource matching, while low-priority orders can only use the remaining resources, thus reflecting the fairness principle of "first come, first served" based on the global optimization of the system.

[0073] 1. Order priority evaluation mechanism based on preset rules

[0074] The present invention proposes a method for prioritizing orders based on order waiting time, and supports flexible configuration of priority factors (time granularity control parameter Δt f ) to achieve flexible differentiated grouping and scheduling of orders.

[0075] 2. Weighted Bipartite Graph Matching Modeling Method with Priority Sorting

[0076] A weighted bipartite graph matching model between vehicles and orders is constructed, and pick-up distance, time cost, and energy consumption are introduced as edge weight components. The optimal order-vehicle matching decision is achieved by minimizing the comprehensive matching cost.

[0077] 3. Priority-layered, step-by-step matching solution strategy

[0078] Based on the order priority queue, the Kuhn-Munkres (KM) algorithm is called layer by layer for matching and solving. High-priority orders use resources first, and low-priority orders are supplemented and matched with the remaining resources to ensure the unity of fairness and efficiency.

[0079] 4. Rolling optimization and real-time update mechanism of the dispatch system

[0080] Establish a scheduling mechanism based on fixed-cycle rolling triggering, extract order and vehicle status in real time during the scheduling cycle, rebuild the priority queue and matching graph structure, and realize dynamic adjustment and continuous optimization of the platform scheduling strategy.

[0081] Compared with the prior art, the present invention has the following beneficial effects:

[0082] 1. Improve matching fairness: The traditional KM algorithm aims to maximize the matching weight, which may cause orders with long waiting time but low edge weight to be ignored for a long time, affecting user satisfaction. This invention introduces "first come, first served"

[0083] The First-Come-First-Served (FCFS) mechanism dynamically divides priority subsets according to the user's waiting time, and gives priority to orders with longer waiting times during the order dispatch process, effectively ensuring service fairness at the user level and improving the overall service experience of the platform.

[0084] 2. Enhanced controllability and flexible scheduling capabilities: The time granularity parameter Δt designed in this invention f As the execution degree regulator of the FCFS principle, it can flexibly control the scheduling trade-off between “fairness first” and “efficiency first”. f When it is smaller, orders will be divided more finely, the FCFS mechanism will be more effective, and matching will tend to prioritize responding to early waiting users; when When , all orders are considered equal, and the system reverts to the standard KM global optimal matching model. The introduction of this parameter enables the platform to flexibly adjust order dispatch strategies based on actual operating strategies (such as emphasizing matching efficiency during peak periods and pursuing maximum revenue during off-peak periods).

[0085] 3. Improve algorithm scalability: In large-scale vehicle and order scenarios, the theoretical time complexity of the KM algorithm is O(n³). Directly matching all orders will result in high computational overhead, affecting real-time performance. This method divides orders into multiple priority subsets and calls the KM algorithm in a hierarchical and progressive manner, significantly reducing the scale of each solution. This improves operational efficiency while ensuring matching quality. In addition, this method has good parallelization potential and can further improve solution speed in multi-threaded or multi-core environments, making it suitable for the deployment of large-scale city-level online ride-hailing dispatch systems.

[0086] 4. Enhance the scalability and adaptability of the model: It should be noted that the order prioritization mechanism proposed in this invention is not limited to being based on "waiting time." The platform can freely select or combine multiple classification criteria based on operational strategies, such as user historical scores, estimated order value, user age group, membership level, destination scarcity, etc., to construct a multi-factor priority scoring function, and then divide and prioritize order subsets based on the score range, thereby adapting to diverse scenarios without changing the core algorithm structure. This highly modular and configurable priority mechanism gives the present invention good scalability and cross-scenario adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 Flow chart of the method of the present invention;

[0088] Figure 2 Modeling a weighted bipartite graph for ride-hailing dispatches. DETAILED DESCRIPTION

[0089] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be further described below.

[0090] Step 1: Extract the orders to be matched from the current system order pool

[0091] Step 1 of the method described herein involves extracting pending orders from the current system's order pool. The specific implementation process is as follows: First, a real-time order management module is deployed within the dispatch system. This module connects to the ride-hailing platform's order database via standard data interface technologies (such as REST APIs, message queues, or database query interfaces). This module periodically or in real time extracts all user orders currently in the "pending" state, forming a set of pending orders. The extracted order information includes detailed information such as the order number, user location (latitude and longitude coordinates), order creation time, current wait time, user rating, and estimated cost. This order information is stored in a high-speed cache database (such as Redis) within the dispatch system in a structured data format (such as JSON or database records) for rapid access during subsequent order prioritization and vehicle order matching steps. Before each dispatch task is executed, the system automatically updates and maintains the cached order information in real time, deleting or updating orders that have been matched, timed out, or canceled. This ensures that the pending order set remains accurate, timely, and effective, providing high-quality data support for subsequent order grouping, prioritization, and optimized matching.

[0092] Step 2: Generate order priority queue based on preset rules

[0093] First, the order dispatching system processes the set of orders to be matched obtained in step 1. The order dispatch system prioritizes and ranks each order. Specifically, the dispatch system includes an order priority assessment module that evaluates each order based on predefined rules and assigns a corresponding priority tag. Priority assessment rules include, but are not limited to, order waiting time, estimated order cost, historical user ratings, and the maximum acceptable waiting time for users.

[0094] Taking the order priority division based on the user waiting time as an example, in order to clarify the specific order priority division process, the present invention introduces a time granularity control parameter Δt f , used to segment the waiting time of an order. The order queue is divided according to the following method:

[0095]

[0096] Where, Indicates order o i Current waiting time, The maximum waiting time for an order allowed by the user. According to the above method, the order collection Divide into K subsets, each subset is assigned a priority label It can be seen from this that the larger the value of the priority label of the order subset, the longer the waiting time of the orders in the subset, the higher the priority, and the priority will be given to processing and matching in the subsequent order dispatch process.

[0097] Finally, the system calculates the subset priority label value ρ k Sort all order subsets to form a complete order priority queue The subset at the head of the queue has the highest priority, while the subset at the tail has the lowest priority. The order priority queue generation process described above is automatically completed by the order priority evaluation module within the dispatch system, and the results are stored in the system's in-memory database or cache module for rapid recall by subsequent vehicle order optimization and matching decisions. The order priority queue generated through this step effectively achieves differentiated and equitable processing of user orders, ensuring that user orders with long wait times or higher order values ​​receive higher priority responses and processing, thereby improving user satisfaction and the overall operational efficiency of the platform.

[0098] It is worth noting that the time granularity control parameter Δt f , as the execution intensity adjustment factor of the "First Come First Served (FCFS)" principle, further enhances the flexibility of system policy regulation. Specifically, Δt f The smaller the value, the more subdivided the orders will be. The longer the waiting time, the higher the priority of the subset. The more obvious the FCFS principle is. When the system is in place, all orders are treated as a whole, and the dispatch process no longer considers time sequence, returning to a scheduling strategy based purely on matching weights. This mechanism not only provides a flexible means of adjusting fairness and efficiency, but also enables the invention to adapt to the differentiated scheduling needs of different platforms and scenarios (such as peak / off-peak hours, fast-paced cities / slow-paced small towns, etc.), with good configurability and practicality.

[0099] Step 3: Build a vehicle-order weighted bipartite graph matching model

[0100] After clarifying order priorities, this step aims to build a matching model based on a weighted bipartite graph to achieve optimal allocation of vehicle resources and user orders. This model quantifies the matching cost between different vehicles and orders and optimizes the matching process with the goal of minimizing overall cost.

[0101] At each scheduling time t, the set of vehicles currently available for receiving orders is recorded as The set of user orders to be served is denoted as Ot ={o1, o2, ..., o n}, to model the matching relationship between vehicles and orders, a weighted bipartite graph G = (X∪Y, E) is constructed, where the node set X = {x1, x2, ..., x m} corresponds to the available vehicles, the node set Y = {y1, y2, ..., y n}Corresponding to the order to be matched, the edge set E in the figure represents the reachable or serviceable relationship between the vehicle and the order. As shown in the figure, each edge in the bipartite graph is accompanied by a matching cost weight The weight represents the comprehensive cost required to assign vehicle v to order o.

[0102] In this invention, the matching cost of each edge is Consider the following three factors:

[0103] Pick-up distance The distance traveled by vehicle v to the location of the passenger of order o (unit: kilometers);

[0104] Pick-up time The time required for the vehicle to reach the order passenger's location (unit: minutes);

[0105] Energy consumption The amount of electricity required by the vehicle to complete the pick-up process (unit: kWh).

[0106] The edge weight is defined as:

[0107]

[0108] Among them, α1, β1, and γ1 are the weight coefficients of the three indicators, which are set according to platform strategy or business preference and are used to balance the impact of different factors on total cost.

[0109] Therefore, the specific objective function is to minimize the overall order-taking cost of the fleet:

[0110]

[0111] in is a decision variable, indicating whether order o is served by vehicle v at time t, and its value is 0 or 1.

[0112] To ensure the feasibility of the matching scheme, this model needs to meet the following constraints.

[0113] (1) Each vehicle can only serve at most one order at a time;

[0114]

[0115] (2) Each order can only be served by one vehicle at a time;

[0116]

[0117] (3) If the order waiting time exceeds the maximum allowed value, the match will be automatically cancelled. large It is a sufficiently large number used to construct the conditional judgment of the big M method.

[0118]

[0119] (4) The vehicle battery power must meet the requirement that the remaining mileage after completing the order service is not less than the preset minimum mileage requirement to ensure that the vehicle can reach the nearest charging station. large is a sufficiently large number used to construct the conditional judgment of the large M method; d min The minimum driving distance that the vehicle must maintain after completing the order. is the pick-up mileage, d o The mileage between the start and end points of the order. is the current remaining driving range of vehicle v.

[0120]

[0121] (5) Binary constraints on decision variables:

[0122]

[0123] Step 4: KM algorithm priority level by level

[0124] After the vehicle-order weighted bipartite graph is constructed in step 3, this step combines the order priority queue generated in step 2. The Kuhn-Munkres algorithm (KM algorithm) is used to solve the order subsets of different priorities in turn, so as to achieve the optimal matching that takes into account both fairness and efficiency of order dispatching.

[0125] The KM algorithm is a classic algorithm for solving the maximum weight matching problem on weighted bipartite graphs. Its basic idea is to construct vertex labels (vertex scores) and augmenting paths, iteratively adjust the vertex labels, and ultimately find a globally optimal solution. Its time complexity is approximately O(n³), and it has good performance for medium-sized tasks.

[0126] In the present invention, the dispatch system is based on the generated order priority queue For each priority subset The KM algorithm is called in order from high priority (long waiting time) to low priority for matching. The specific process is as follows:

[0127] 1. Initialize the resource collection:

[0128] Currently unassigned vehicle collection:

[0129] The current priority level index is processed from k=K to k=0 in sequence

[0130] 2. Priority subgraph extraction:

[0131] For the current priority order subset Select vehicles that can accept orders from the remaining vehicle set and construct a local subgraph The edge weight is defined in step 3

[0132] 3. KM algorithm call:

[0133] In each priority subgraph G k On the top, call the KM algorithm to solve the minimum cost matching (or maximum negative weight matching)

[0134] 4. Results update:

[0135] The current matching result M k Merge into the total matching result set M←M∪M k

[0136] Remove the vehicle that has participated in the match from the remaining vehicle set: V t，remain ←V t，remain \Dom(M k )

[0137] 5. Iterate layer by layer until the lowest priority is processed

[0138] Through the above mechanism, high-priority orders are given priority in resource matching, while low-priority orders can only use the remaining resources, thus reflecting the fairness principle of "first come, first served" based on the global optimization of the system.

[0139] Step 5: Rolling update execution of dispatch system

[0140] The dispatch system employs a fixed-cycle rolling execution strategy, continuously repeating the entire "extraction-sorting-mapping-matching-dispatch-update" process at a fixed interval (e.g., 3 minutes), forming a closed-loop scheduling system. During each dispatch cycle, only pending orders and available idle vehicles in the current order pool are processed. This enables continuous, high-frequency, and low-latency dispatch optimization, meeting the comprehensive requirements of real-time performance, stability, and responsiveness.

[0141] The above description is merely a preferred embodiment of the present invention and does not limit the present invention in any way. Any person skilled in the art who, without departing from the scope of the present invention, makes any equivalent substitution, modification, or other changes to the technical solution and technical content disclosed in the present invention shall be deemed to be within the scope of the present invention and still fall within the scope of protection of the present invention.

Claims

1. A method for dispatching online ride-hailing orders integrated with order priority sorting, characterized in that: The steps include: Step 1: Extract the orders to be matched from the current system order pool; Step 2: Generate order priority queue based on preset rules; Step 3: Build a vehicle-order weighted bipartite graph matching model; Step 4: KM algorithm priority is solved step by step; Step 5: Rolling update of the dispatch system is executed.

2. The method for dispatching online ride-hailing orders based on order priority sorting according to claim 1 is characterized in that: The method of step one: a real-time order management module is set up inside the dispatch system, and the module is connected to the online car-hailing platform order database through standard data interface technology, from which all user orders currently in the waiting state are periodically or in real time extracted to form a set of orders to be matched.

3. The method for dispatching online ride-hailing orders based on order priority sorting according to claim 2 is characterized in that: Order information is stored in a structured data format in a high-speed cache database within the dispatch system.

4. The method for dispatching online ride-hailing orders integrated with order priority sorting according to claim 1, characterized in that: The method of step 2: the order dispatching system obtains the set of orders to be matched To evaluate and sort the priority of each order in the dispatch system, an order priority evaluation module is set up inside the dispatch system. This module evaluates each order based on pre-defined rules and assigns corresponding priority labels.

5. The method for dispatching online ride-hailing orders based on order priority sorting according to claim 4 is characterized in that: Priority evaluation rules include: order waiting time, estimated order cost, user historical evaluation level, and the maximum waiting time acceptable to the user.

6. The method for dispatching online ride-hailing orders based on order priority sorting according to claim 5 is characterized in that: The method of prioritizing orders based on the waiting time of the order: In order to clarify the specific order priority division process, a time granularity control parameter Δt is introduced f , which is used to segment the waiting time of orders and divide the order queue according to the following method: Where, Indicates order o i Current waiting time, The maximum waiting time for an order allowed by the user, according to the above method, the order collection Divide into K subsets, each subset is assigned a priority label It can be seen that the larger the priority tag value of the order subset, the longer the waiting time of the orders in the subset, the higher the priority, and the priority will be given priority in the subsequent order dispatch process. Finally, the system will prioritize the processing and matching based on the subset priority tag value ρ. k Sort all order subsets to form a complete order priority queue The priority of the head subset of the queue is the highest, and the priority of the tail subset is the lowest. The generation process of the above order priority queue is automatically completed by the order priority evaluation module in the dispatch system, and the results are stored in the system memory database or cache module for rapid call by subsequent vehicle order optimization matching decisions.

7. The method for dispatching online ride-hailing orders integrated with order priority sorting according to claim 1, characterized in that: The method of step 3: At each scheduling time t, the set of vehicles currently available for receiving orders is recorded as The set of user orders to be served is denoted as O t ={o1, o2, ..., o n }, to model the matching relationship between vehicles and orders, a weighted bipartite graph G = (X∪Y, E) is constructed, where the node set X = {x1, x2, ..., x m } corresponds to the available vehicles, the node set Y = {y1, y2, ..., y n } corresponds to the order to be matched, the edge set E represents the reachable or serviceable relationship between the vehicle and the order, and the matching cost weight represents the comprehensive cost required to assign vehicle v to order o, The matching cost of each edge Consider the following three factors: Pick-up distance The distance traveled by vehicle v to the location of the passenger of order o (unit: kilometers); Pick-up time The time required for the vehicle to reach the order passenger's location (unit: minutes); Energy consumption The amount of electricity required by the vehicle to complete the pick-up process (unit: kWh), The edge weight is defined as: Among them, α1, β1, and γ1 are the weight coefficients of the three indicators, which are set according to the platform strategy or business preference to balance the impact of different factors on the total cost. Therefore, the specific objective function is to minimize the overall order-taking cost of the fleet: in is a decision variable, indicating whether the order o is served by vehicle v at time t, and takes the value of 0 or 1. To ensure the feasibility of the matching scheme, this model needs to meet the following constraints: (1) Each vehicle can only serve at most one order at a time; (2) Each order can only be served by one vehicle at a time; (3) If the order waiting time exceeds the maximum allowed value, the match will be automatically cancelled, where M large is a sufficiently large number used to construct the conditional judgment of the big M method. (4) The vehicle battery power must meet the requirement that the remaining mileage after completing the order service is not less than the preset minimum mileage requirement to ensure that the vehicle can reach the nearest charging station, where M large is a sufficiently large number used to construct the conditional judgment of the large M method; d min The minimum driving distance that the vehicle must maintain after completing the order. is the pick-up mileage, d o is the mileage between the start and end point of the order, is the current remaining driving range of vehicle v, (5) Binary constraints on decision variables:

8. The method for dispatching online ride-hailing orders based on order priority sorting according to claim 7 is characterized in that: After completing the vehicle-order weighted bipartite graph, the order priority queue generated by combining The Kuhn-Munkres algorithm (KM algorithm) is used to solve the order subsets of different priorities in turn to achieve the optimal matching that takes into account both fairness and efficiency of order dispatching.

9. The method for dispatching online ride-hailing orders based on order priority sorting according to claim 8 is characterized in that: The dispatch system is based on the generated order priority queue For each priority subset The KM algorithm is called in order from high priority, long waiting time, to low priority for matching. The specific process is as follows: (1) Initialize resource collection: Currently unassigned vehicle collection: The current priority level index is processed from k=K to k=0 in sequence (2) Priority subgraph extraction: For the current priority order subset Select vehicles that can accept orders from the remaining vehicle set and construct a local subgraph The edge weight is defined in step 3 (3) KM algorithm call: In each priority subgraph G k On the top, call the KM algorithm to solve the minimum cost matching (or maximum negative weight matching) (4) Result update: The current matching result M k Merge into the total matching result set M←M∪M k Remove the vehicle that has participated in the match from the remaining vehicle set: V t，remain ←V t，remain \Dom(M k ) (5) Iterate layer by layer until the lowest priority level is processed Through the above mechanism, high-priority orders are given priority in resource matching, while low-priority orders can only use the remaining resources, thus reflecting the fairness principle of "first come, first served" based on the global optimization of the system.

Images